No association between fiber intake and prostate ... - Semantic Scholar

0 downloads 0 Views 790KB Size Report
mary OR for prostate cancer ranged from 0.87 (0.75 to. 0.99) when the ..... Lewis JE, Soler-Vila H, Clark PE, Kresty LA, Allen GO, Hu JJ. Intake of ... Higgins JA.
Sheng et al. World Journal of Surgical Oncology (2015) 13:264 DOI 10.1186/s12957-015-0681-8

WORLD JOURNAL OF SURGICAL ONCOLOGY

RESEARCH

Open Access

No association between fiber intake and prostate cancer risk: a meta-analysis of epidemiological studies Tao Sheng1, Rui-lin Shen1, Huan Shao1 and Tian-hong Ma2*

Abstract Background: The findings of epidemiologic studies on the association between fiber intake and prostate cancer risk remain conflicting. We aimed to examine this association by conducting a meta-analysis of epidemiological studies. Methods: Relevant studies were identified by PubMed (1966 to March 2015) and Embase (1974 to March 2015) database search through March 2015. We included epidemiological studies that reported relative risks (RRs) or odds ratios (ORs) with 95 % confidence intervals (CIs) for the association between dietary fiber intake and prostate cancer risk. Random effects models were used to calculate the summary risk estimates. Results: For the highest compared with the lowest dietary fiber intake, a significantly decreased risk with prostate cancer was observed in case-control studies (OR = 0.82; 95 % CI, 0.68–0.96), but not in cohort studies (RR = 0.94; 95 % CI, 0.77–1.11). The combined risk estimate of all studies was 0.89 (95 % CI, 0.77, 1.01). A significant heterogeneity was observed across studies (p = 0.005). There was no evidence of significant publication bias based on Begg’s funnel plot (p = 0.753) or Egger’s test (p = 0.946). Conclusions: This meta-analysis suggests the absence of evidence for association between dietary fiber intake and prostate cancer risk. Keywords: Prostatic neoplasms, Dietary fiber, Meta-analysis, Epidemiology

Background Prostate cancer is the second most common cancer among men in the world, with 1.1 million new cases diagnosed in 2012 worldwide, accounting for about 7.9 % of all cases of cancer [1]. The high prevalence and incidence of prostate cancer have resulted in a large public health burden. Age and family history are well-established and strong risk factors for prostate cancer [2]. Environmental factors such as diet are believed to play an important role in the prevention of prostate cancer because of the wide international variation in incidence [3]. Although dietary factors have long been suspected to be implicated in the development of prostate cancer, no major modifiable risk factor has been established. During

the last few years, increased intake of dietary fibers has been associated with decreased risk of several cancers, such as colorectal, breast, ovarian, and upper aerodigestive tract cancers [4–7]. However, results from epidemiological studies regarding prostate cancer are sparse and inconsistent. The 2007 World Cancer Research Fund (WCRF) Second Expert Report concluded that the data were too inconsistent to draw a conclusion on the association between dietary fiber intake and prostate cancer risk [8]. Since that report was released, five prospective studies have been published on this association [9–13]. To quantitatively assess the accumulated evidence for a role of dietary fiber consumption on prostate cancer risk, we carried out a systematic review and meta-analysis of published epidemiological studies.

* Correspondence: [email protected] 2 Department of Pharmacy, Jiaxing Affilated Hospital of Zhejiang Chinese Medical University, Zhongshan East Road 1501, Jiaxing, Zhejiang Province 314001, China Full list of author information is available at the end of the article © 2015 Sheng et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Sheng et al. World Journal of Surgical Oncology (2015) 13:264

Methods Selection of studies

Two authors performed a computerized blinded search of MEDLINE (1966 to March 2015) and Embase (1974 to March 2015) databases for relevant epidemiologic studies of dietary fiber consumption in relation to the risk of prostate cancer published in English. Additional publications identified by hand-searching of references of retrieved articles were also included. For computer searches, we used the following words in any field: “fiber” or “fibre” combined with “prostate carcinoma” or “prostatic cancer” or “prostate cancer” or “prostatic carcinoma”. Studies were included in the meta-analyses if they presented estimates of the odds ratio (OR) or relative risk (RR) and the corresponding confidence interval (CI) from a case-control or cohort study on the association between fiber intake and incidence of prostate cancer. When multiple reports were published on the same study population, we included the study with the largest number of cases. Figure 1 gives the flowchart for selection of articles. The primary literature search identified 505 records. After screening the titles and abstracts, 486 articles were excluded because they were either duplicates, review articles, or irrelevant to the current study. Nineteen full-text papers were retrieved. In addition, we included ten studies after reviewing reference lists of retrieved articles or preceding reviews. Twelve studies [14–25] were excluded mostly because of insufficient information to compute its RR or OR and 95 % CI. Finally, we identified 5 prospective studies [9–13] and 12 case-control studies [26–37] with data that were eligible for inclusion in the meta-analysis.

Page 2 of 11

of publication, the country in which the study was conducted, study design, year of follow-up (cohort studies), year of data collection (case-control studies), sample size, evaluation of exposures, the RR or OR and its 95 % CIs, exposure assessment and range of exposure, and adjusted covariates. Data extraction was conducted independently by two authors, with disagreements resolved by consensus. Considering that prostate cancer is a relatively rare disease, the RR was assumed approximately the same as OR, and the OR was used as the study outcome. If a study provided several ORs, we extracted the ORs reflecting the greatest degree of control for potential confounders. Oishi et al. [26] presented two ORs for benign prostatic hyperplasia (BPH) and hospital controls, respectively. We chose the risk estimate comparing prostate cancer with hospital controls instead of BPH because it may increase the chance of diagnosing an incidental prostate cancer [38]. Quality assessment

The study quality was assessed using the nine-star Newcastle-Ottawa Scale (The Newcastle-Ottawa Scale for assessing the quality of nonrandomized studies in metaanalyses. Ottawa, Canada: Dept of Epidemiology and Community Medicine, University of Ottawa. http:// www.ohri.ca/programs/clinical_epidemiology/oxford.htm). NOS is an eight-item instrument that allows for the assessment of the patient selection, study comparability, and exposure (for case-control study) or outcome (for cohort study). The range of possible scores is 0–9. The study with score more than 6 was considered of high quality.

Data extraction and classification

Statistical analysis

The following pieces of information were extracted from published studies: the name of the first author, the year

We used random effects models to calculate summary ORs and 95 % CIs for the highest vs. the lowest levels of

Fig. 1 Flowchart of study selection

Sheng et al. World Journal of Surgical Oncology (2015) 13:264

dietary fiber because it used a combination of withinstudy variance and between-study variance for computing weights. We evaluated the heterogeneity among studies with the Cochrane Q test [39] and I2 score [40]. We also estimated the 95 % prediction interval, which further accounts for between-study heterogeneity and evaluates the uncertainty for the effect that would be expected in a new study addressing that same association [41]. To explore the sources of heterogeneity across studies, subgroup analyses were conducted according to study design, study quality, geographic region, and method of dietary assessment. Because adjustments for confounding factors were not consistent between the studies, we also conducted the subgroup analysis according to whether the risk estimates had been adjusted for family history of prostate cancer, body mass index (BMI), and total energy intake. In addition, we further performed a sensitivity analysis to explore sources of heterogeneity. Each study was omitted at a time to assess robustness of the results. In addition to those methods, the Galbraith plot was also used to detect the possible sources of heterogeneity, and a re-analysis was conducted with exclusion of the studies possibly causing the heterogeneity. Meta-regression was also applied to measure the subgroup interaction. The p value for interaction between two groups is the comparison of subgroup vs. the other. We used p < 0.10 as the indicator of significant interaction. Publication bias was assessed by Begg’s [42] and Egger’s [43] test. All analyses were performed by using STATA version 11.0 (StataCorp). A p value < 0.05 was considered significant.

Results The characteristics of these studies and the variables evaluated are listed in Table 1. Six studies were conducted in North America [10, 31, 33, 34, 36, 37], seven in Europe [9, 11, 12, 28, 30, 32, 35], two in Japan [13, 26], one in South Africa [27], and one in Uruguay [29]. Overall, this meta-analysis included more than 8000 cases of prostate cancer. Information on fiber intake was obtained by interview or self-administered questionnaire using food frequency questionnaires (FFQ) except one using 24-h dietary record [12]. All of the included studies adjusted for age, and 14 of them included adjustment for energy intake [6, 9–13, 28–31, 33–36], 8 adjusted for family history [6, 10, 12, 29–31, 36, 37], and 8 adjusted for BMI [10, 12, 13, 29, 30, 33, 36, 37]. As shown in Fig. 2, a statistically significant protective effect of dietary fiber intake on prostate was observed in case-control studies (OR = 0.82; 95 % CI, 0.68–0.96), while no such effect was observed in cohort studies (RR = 0.94; 95 % CI, 0.77–1.11). There was no evidence of heterogeneity among case-control (p = 0.277, I2 = 17 %), but significant heterogeneity among cohort studies (p = 0.004, I2 = 74.3 %). When all these studies

Page 3 of 11

were analyzed together, no association was observed between fiber intake and risk of prostate cancer (summary OR = 0.89; 95 % CI, 0.77–1.01), with significant heterogeneity among studies (p = 0.005, I2 = 53.6 %). The wide 95 % prediction interval also included the null value and reflected the significant heterogeneity (0.59, 1.52). In a sensitivity analysis excluding one study at a time, the summary OR for prostate cancer ranged from 0.87 (0.75 to 0.99) when the study by Drake et al. [11] was excluded to 0.93 (0.83 to 1.03) when the study by Deschasaux et al. [12] was excluded. Through the Galbraith plot, four studies were identified as the major sources of heterogeneity (Fig. 3). After excluding these four studies, there was no study heterogeneity (p = 0.915, I2 = 0), and the overall association turned out to be null (OR 1.00, 95 % CI 0.93–1.07). There was no evidence of significant publication bias either with the Egger’s test (p = 0.946) or Begg’s funnel plot (p = 0.753) (Fig. 4). Next, we performed subgroup analyses by study quality, geographical region, and the method of exposure assessment (Table 2). When we stratified by study quality, more significant association was observed in studies of low-quality (OR 0.73, 95 % CI 0.56–0.90) compared with studies of high-quality (OR 0.96, 95 % CI 0.83–1.08). Considering the geographic area, the pooled OR was 0.90 (95 % CI, 0.65–1.16) in European studies, 0.90 (95 % CI 0.64–1.06) in North American studies, and 0.95 (95 % CI, 0.72–1.17) in Japanese studies. When separately analyzed by exposure assessment, the ORs were 0.93 (95 % CI 0.76–1.09) for studies that used an interview and 0.94 (0.76–1.10) for with a self-administered questionnaire, respectively. We also investigated the impact of some confounding factors on the estimates of ORs (Table 2). Family history is the established risk factor for prostate cancer; BMI and energy are potential confounders of the relationship between fiber intake and the risk of prostate cancer. We found that the non-significant relationships between prostate cancer and fiber intake were consistent in all subgroups, whether controlled for family history, BMI, and energy intake or not. Moreover, six studies in our analysis adjusted for these three confounders simultaneously. Therefore, we examined whether more thoroughly adjusting for potential confounders affected the pooled OR. The effect estimates for studies that adjusted for these three confounders or not were ORs of 0.82 (95 % CI 0.54–1.09) and 0.95 (0.84–1.05), respectively. In addition, after stratification according to food source and solubility, none of the subtypes could lower the incidence of prostate cancer significantly, except for legume fiber, though it is based on only one cohort study [12]. We also pooled the ORs by clinical characteristics of prostate cancer. The summary ORs did not indicate that high fiber intake had a significant

Study Country design

Study period Cases/subjects Exposure range

RR (95 % CI)

Variables of adjustment

Study Other variables evaluated qualitya

Assessment

Oishi et al. 1988 [26]

HCC

Japan

1981–1984

100/200

Ever vs. none

0.78 (0.45–1.37)

Age

5

None

Interview FFQ (31 items)

Walker et al. 1992 [27]

PCC

South Africa 1998–1990

166/332

≥15 vs.